3 ms·
The mitigations they're talking about bypassing are, like all mitigation techniques, intended to ensure that if someone has a security hole they cannot make use
by makomk 12y ago
The mitigations they're talking about bypassing are, like all mitigation techniques, intended to ensure that if someone has a security hole they cannot make use of it. That's their entire point; if there were no security holes, there'd be no need for stuff like DEP, SMEP, and ASLR in the first place. In practice that's not a realistic assumption, which is why companies like Intel have gone to the trouble of adding hardware features that have zero benefit unless there's a security hole in the operating system.
- acqq 12y agoDEP, SMEP, ASLR are there to make the user-mode exploits harder. But if that bypass depends on having the "root" "write anywhere" hole, well, it uses the quite a big hole, and they just demonstrate that it is actually a big one. As far as I undestand, DEP and SMEP simply use the VM page flags controlled by the "root." The article authors depend on having the "root" write access to the given flags, and then they claim that they "bypass" DEP and SMEP, whereas they actually change the mentioned flags. They also explain that as soon as they have that, they simply aren't constrained by the ASLR at all.